IARA - Interpretable AI for Risk Assessment
IARA is a next-generation in silico New Approach Methodology (NAM) that combines artificial intelligence with a curated human Biomedical Knowledge Graph (BKG) to predict liver, cardiac, nervous system and kidney toxicity.
📈 Model Performance
| Organ | CV ROC-AUC | Test ROC-AUC | Accuracy | MCC | Sensitivity | Specificity | F1 Score | Youden threshold (J) |
|---|---|---|---|---|---|---|---|---|
| HEART | 0.74 ± 0.02 | 0.79 ± 0.02 | 0.71 ± 0.06 | 0.44 ± 0.10 | 0.69 ± 0.13 | 0.75 ± 0.11 | 0.73 ± 0.12 | 0.75 ± 0.09 |
| LIVER | 0.87 ± 0.01 | 0.84 ± 0.01 | 0.72 ± 0.05 | 0.46 ± 0.06 | 0.78 ± 0.08 | 0.68 ± 0.12 | 0.69 ± 0.03 | 0.35 ± 0.11 |
| NERVOUS SYSTEM | 0.82 ± 0.01 | 0.82 ± 0.01 | 0.73 ± 0.03 | 0.48 ± 0.05 | 0.78 ± 0.08 | 0.69 ± 0.09 | 0.73 ± 0.03 | 0.47 ± 0.10 |
| KIDNEY | 0.77 ± 0.01 | 0.64 ± 0.02 | 0.62 ± 0.03 | 0.23 ± 0.06 | 0.65 ± 0.07 | 0.58 ± 0.08 | 0.65 ± 0.04 | 0.55 ± 0.09 |
Select one of the two available query modes to perform toxicity prediction. Choose between a compound identifier–based workflow or a protein target–based workflow.